DE102017216536B4 - Method for compensating for disturbances in a measured angle signal of a magnetic angle sensor of an electrical machine and a correspondingly designed microcontroller, an electrical machine, and a computer program product - Google Patents

Abstract

Method for compensating for disturbances in a measured angle signal of a magnetic angle sensor (12) of an electrical machine (10), the method having: - receiving a measured angle signal, - estimating a current error and/or an offset error of the measured angle signal, estimating the current error and/or the offset error, the measured angle signal, which comprises at least two components, is transferred to a calculation program stored in a microcontroller,- calculating a rotor angle to be expected from the measured angle signal, taking into account the estimated current error and/or the estimated offset error, during operation the electrical machine (10), characterized in that an amplitude error and/or an offset error is also taken into account to calculate a rotor angle to be expected, in particular is transferred to the calculation program.

Description

The present invention relates to a method for compensating for disturbances in a measured angle signal of a magnetic angle sensor of an electrical machine, a microcontroller for calculating disturbances in a measured angle signal of a magnetic angle sensor of an electrical machine, and a correspondingly designed electrical machine.

In order to control an electrical machine, the angular position of the rotor must be known. Inductive angle sensors or reluctance resolvers are often used for this. However, inductive angle sensors and reluctance resolvers are expensive.

The use of magnetic angle sensors, on the other hand, is inexpensive and also accurate. A magnetic angle sensor includes what is known as an integrated cross-hall element. This cross-hall element generates an electrical parameter in relation to the flow direction of an applied external magnetic field. By changing a magnetic field that acts on the cross Hall element, such as when rotating a magnetic field of a permanent magnet around the central axis of the sensor element, two 90° phase-shifted sine or cosine curves are generated, based on which the rotation angle is calculated as an absolute value can be detected. Evaluation electronics integrated in the magnetic angle sensor convert these measured sine or cosine curve signals into a linear voltage or current signal. Due to the absolute measuring principle, the correct angle of rotation can be output even after a power failure. However, magnetic angle sensors are sensitive to magnetic interference fields, which occur, for example, due to current-carrying conductors and/or due to permanent magnets in electrical machines. When using magnetic angle sensors in electrical machines, methods for compensating for magnetic interference fields are therefore required.

From the DE 10 2015 115 247 A1 a linear position and rotary position magnetic sensor are known. A corresponding system and procedure are also described.

In the U.S. 2006/0 125 139 A1 a synchronous motor drive unit is described.

The DE 10 2011 076 734 A1 relates to a method and apparatus for angle estimation in a synchronous machine.

It is an object of the present invention to provide a method for compensating for interference fields and to provide a correspondingly designed microcontroller and a correspondingly configured electrical machine with which the method for compensating for interference fields can be carried out.

This object is achieved by a method for compensating for interference in a measured angle signal of a magnetic angle sensor of an electrical machine according to claim 1 and by a microcontroller for calculating interference in a measured angle signal of a magnetic angle sensor of an electrical machine according to claim 10. This object is also achieved by an electrical machine with a magnetic angle sensor and such a microcontroller according to claim 11 and by a computer program product according to claim 12, which can execute the method.

• - Empfangen eines gemessenen Winkelsignals,
• - Abschätzen eines Stromfehlers und/oder eines Versatzfehlers des gemessenen Winkelsignals,
• - Berechnen eines zu erwartenden Rotorwinkels aus dem gemessenen Winkelsignal unter Berücksichtigung des abgeschätzten Stromfehlers und/oder des abgeschätzten Versatzfehlers, insbesondere während des Betriebes der elektrischen Maschine.

According to the present invention, we therefore propose a method for compensating for interference in a measured angle signal of a magnetic angle sensor of an electrical machine, the method having the following steps:

• - receiving a measured angle signal,
• - estimating a current error and/or an offset error of the measured angle signal,
• - Calculating a rotor angle to be expected from the measured angle signal, taking into account the estimated current error and/or the estimated offset error, in particular during operation of the electrical machine.

$${\text{U}}_{\text{cos}}=\text{A cos }\left(2\mathrm{\alpha }\right)$$

According to the proposal, a changing magnetic field is measured using a magnetic angle sensor. By rotating a rotor in an electrical machine, on which a permanent magnet is arranged, a magnetic field that changes over time can be generated in the angle sensor, for example, or a direction-dependent, changing magnetization can be detected in the angle sensor. The magnetic angle sensor preferably has two Wheatstone bridges, in particular at an angle of 45° to one another, so that the angle sensor measures two voltage signals according to equations (1) and (2): $${\text{u}}_{\text{sin}}=\text{A sin}\left(2\mathrm{a}\right)$$

$${\text{u}}_{\text{cos}}=\text{A cos}\left(2\mathrm{a}\right)$$

$$U_{cos}=A_{cos}\cdot \text{cos}\left(2\alpha \right).$$

An angle sensor makes use of the so-called magnetoresistive effect, which occurs, for example, in the Permalloy alloy. With this effect, the resistance changes depending on the direction of the changing, induced magnetic field. However, the measured sensor signal may have a current error and/or an offset error due to current disturbances or due to an imprecise placement of the permanent magnet at one end of the rotor. In the event of a current error, the measured phase currents themselves are usually not faulty. However, the magnetic field caused by the phase currents causes an error in the sensor, which is compensated with the proposed method. In general, the sensor signal can be disturbed by various errors, which means that the amplitudes A _sin , A _cos of the two component sensor signals can be different, namely: $$u_{sin}=A_{sin}\cdot \text{sin}\left(2a\right)$$

$$u_{cOs}=A_{cOs}\cdot \text{cos}\left(2a\right).$$

$$U_{cos}=A_{cos}\cdot \text{cos}\left(2\alpha \right)+O_{cos}.$$

For example, both signals can have an offset error O _sin , O _cos : $$u_{sin}=A_{sin}\cdot \text{sin}\left(2a\right)+O_{sin}$$

$$u_{cOs}=A_{cOs}\cdot \text{cos}\left(2a\right)+O_{cOs}.$$

$$U_{cos}=A_{cos}\cdot \text{cos}\left(2\alpha +\epsilon \right)+O_{cos}.$$

Furthermore, both signals can have an offset error ε, for example, if the sensor is not centered on the rotor of the electrical machine: $$u_{sin}=A_{sin}\cdot \text{sin}\left(2a+e\right)+O_{sin}$$

$$u_{cOs}=A_{cOs}\cdot \text{cos}\left(2a+e\right)+O_{cOs}.$$

$$U_{cos}=A_{cos}\cdot \text{cos}\left(2{\alpha }_I+\epsilon \right)+O_{sin},$$

wobei a_I ein Winkel des magnetischen Feldes ist, welches durch einen Phasenstromfehler gestört ist. Ein Ziel der vorliegenden Erfindung ist es, möglichst alle Fehler zu kompensieren.Furthermore, when electric currents flow in the rotor, the measured magnetic field, which is measured by the sensor, can be disturbed. The sensor signal, which includes all possible errors, is then given by: $$u_{sin}=A_{sin}\cdot \text{sin}\left(2a_I+e\right)+O_{sin}$$

$$u_{cOs}=A_{cOs}\cdot \text{cos}\left(2a_I+e\right)+O_{sin},$$

where a _I is an angle of the magnetic field perturbed by a phase current error. One aim of the present invention is to compensate for as many errors as possible.

According to the proposal, the current error and/or the offset error are taken into account in the proposed method, in addition to other errors, by estimating a size of one or both of these errors and correcting the measured angle signal by such an error. After the measured angle signal has been corrected for such an error, an actual rotor angle to be expected can be calculated from the measured angle signal. Such a calculation of the angle signal to be expected, taking into account the estimated current error and/or the estimated offset error, can be carried out during operation of the electrical machine. With the proposed method, which corresponds to the functioning of a digital filter, an actual angular position of the rotor can consequently be determined. Digital filters are used in control engineering to manipulate a signal. In contrast to analog filters, no electronic components are used for a digital filter but logic components. In addition, a digital filter processes signals that are discrete in terms of time and value, such as those recorded by a magnetic angle sensor. With the proposed method, the error in the angular position of the rotor at a speed of, for example, up to 18000 rpm is preferably less than or equal to 1° degree. Furthermore, an angular velocity of the rotor can be determined with the proposed method. For example, during constant acceleration or deceleration of up to 30000 rpm/s, the angular velocity error can be less than or equal to 10 rpm. A measurement error of up to 20% can advantageously be compensated for with the proposed method, so that an actual, expected rotor angle can be determined with the proposed method. In addition to the current errors and/or displacement errors mentioned, the measurement errors can also be an offset error. The proposed method can be used, for example, to compensate for an offset error that is caused by a radial misalignment of up to 1 mm and an axial misalignment of up to 5 mm of the permanent magnet on the rotor. Furthermore, a current error can be compensated by up to 400A. In addition, the proposed method can be carried out quickly, in particular with a measurement time of 100 μs per measurement.

According to the invention, the measured angle signal, which comprises at least two components, is transferred to a calculation program stored in a microcontroller in order to estimate the current error and/or the offset error.

zuordenbar ist, welche/welcher durch folgendes Gleichungssystem darstellbar ist: $$\frac{d}{dt}\underset{\xi }{\underbrace{\left(\begin{array}{c}\ddot{\alpha }\\ \dot{\alpha }\\ \alpha \end{array}\right)}}=\underset{A}{\underbrace{\left[\begin{array}{ccc}0& 0& 0\\ 1& 0& 0\\ 0& 1& 0\end{array}\right]}}\cdot \underset{\xi }{\underbrace{\left(\begin{array}{c}\ddot{\alpha }\\ \dot{\alpha }\\ \alpha \end{array}\right)}}+\underset{B}{\underbrace{\left[\begin{array}{c}1\\ 0\\ 0\end{array}\right]}}\cdot \tau$$

$$\frac{d}{dt}\xi =A\cdot \xi +B\cdot \tau .$$

The calculation program is software that can solve systems of equations, ie matrices. One aim of the method described is to estimate or calculate the rotor position angle by evaluating the measured angle sensor signal. When using the calculation program are preferred no assumptions are made about a mass, in particular an author or a power train of a car, or an influence of a current on the movement of the rotor. Instead, it is assumed that the rotor has an angular acceleration, an angular velocity and an angle, i.e. a state $\left(\begin{array}{c}\ddot{a}\\ \dot{a}\\ a\end{array}\right),$

can be assigned, which can be represented by the following system of equations: $$\frac{\mathrm{i.e}}{\mathrm{i.e}t}\underset{\xi }{\underbrace{\left(\begin{array}{c}\ddot{a}\\ \dot{a}\\ a\end{array}\right)}}=\underset{A}{\underbrace{\left[\begin{array}{ccc}0& 0& 0\\ 1& 0& 0\\ 0& 1& 0\end{array}\right]}}\cdot \underset{\xi }{\underbrace{\left(\begin{array}{c}\ddot{a}\\ \dot{a}\\ a\end{array}\right)}}+\underset{B}{\underbrace{\left[\begin{array}{c}1\\ 0\\ 0\end{array}\right]}}\cdot \tau$$

$$\frac{\mathrm{i.e}}{\mathrm{i.e}t}\xi =A\cdot \xi +B\cdot \tau .$$

$$\tilde{\vartheta }:=2\cdot \tilde{\alpha }$$

$$\text{sin}\left(\vartheta -\tilde{\vartheta }\right)=\text{sin}\left(\vartheta \right)-\text{cos}\left(\tilde{\vartheta }\right)-\text{cos}\left(\vartheta \right)\cdot \text{sin}\left(\tilde{\vartheta }\right)$$

$$\text{sin}\left(\vartheta -\tilde{\vartheta }\right)\approx \vartheta -\tilde{\vartheta }\text{ f}\ddot{\text{u}}\text{r kleine Differenzen und damit:}$$

$$\vartheta -\tilde{\vartheta }\approx \text{sin}\left(\vartheta \right)\cdot \text{cos}\tilde{\vartheta }-\text{cos}\left(\vartheta \right)\cdot \text{sin}\left(\tilde{\vartheta }\right)\text{.}$$

Here E is the state of the system and A is the associated state matrix. The variable τ denotes the unknown change in angular acceleration. For an observer we assume τ =0. In order to obtain a structure of the observer, an estimated angle α̃ must be able to be assigned to a measured angle signal. A difference between an estimated angle α̃ and a measured angle signal α can then be described by the following equations: $$\vartheta :=2\cdot a$$

$$\tilde{\vartheta }:=2\cdot \tilde{a}$$

$$\text{sin}\left(\vartheta -\tilde{\vartheta }\right)=\text{sin}\left(\vartheta \right)-\text{cos}\left(\tilde{\vartheta }\right)-\text{cos}\left(\vartheta \right)\cdot \text{sin}\left(\tilde{\vartheta }\right)$$

$$\text{sin}\left(\vartheta -\tilde{\vartheta }\right)\approx \vartheta -\tilde{\vartheta }\text{f}\ddot{\text{and}}\text{r small differences and thus:}$$

$$\vartheta -\tilde{\vartheta }\approx \text{sin}\left(\vartheta \right)\cdot \text{cos}\tilde{\vartheta }-\text{cos}\left(\vartheta \right)\cdot \text{sin}\left(\tilde{\vartheta }\right)\text{.}$$

gegeben ist. Das System (A,C) ist beobachtbar und wird mit dem vorgeschlagenen Verfahren gelöst.The difference α - α̃ can be expressed by the difference ϑ - ϑ̃, so that the sensor signal system through $$\vartheta =\underset{c}{\underbrace{\left[\begin{array}{ccc}0& 0& 2\end{array}\right]}}\cdot \xi$$

given is. The system (A,C) is observable and is solved with the proposed method.

According to a preferred embodiment of the method, the calculation program includes a model with which a rotor state detected by means of the received angle signal comprising an angle and an angular velocity and an angular acceleration, taking into account various errors in the angle signal comprising an amplitude error and/or an offset error and/or the displacement error and /or the current error is mapped to a current actual rotor state.

Different errors can thus be taken into account in the proposed method, or the measured angle signal can be corrected on the basis of different errors. The correction preferably includes adding or subtracting an estimated value from a measured value. It can also include multiplying or dividing or executing a concatenation of the basic arithmetic rules.

Preferably the use of a model related to the current error and/or the offset error is applied. It has been shown here that taking into account a current error and/or an offset error in the model for calculating a rotor angle to be expected leads to correct values for the rotor angle.

An offset error is defined here as a radial offset and/or an axial offset of the permanent magnet relative to an axial axis of the rotor of the electrical machine. Such a radial offset can be up to 2 mm radially, for example. An axial offset can be up to 5mm, for example. Such radial and/or axial offset errors can be taken into account with the proposed method. A power fault is one of the usual uncertainties or momentary power interruptions in an electrical system. A current error of up to 400A can be taken into account with the proposed method.

Preferably, the estimated current error and/or the estimated offset error is/are used to estimate a rotor angle error by determining a difference of a measured rotor angle minus an estimated rotor angle.

The current error and/or the estimated offset error can be taken into account by subtracting an estimated rotor angle from the measured rotor angle. It is therefore assumed that the measured rotor angle turns out to be too large due to the current error and/or the offset error measured at the same time. It is also conceivable for the values to be added instead of subtracted.

Estimation of a current error can be estimated by the following equation, for example. Assuming that the errors in the components x _I and y _I are compensated by the Pha If the currents I ₁ ,I ₂ ,I ₃ are caused and simply add up, the components x _I and y _I can be written as: $$\begin{array}{l}{x}_I=p_{x1}\cdot I_1+p_{x2}\cdot I_2+p_{x3}\cdot I_3\\ y_I={\mathrm{<figure-callout\; id="1"\; label="p\; y"\; filenames="DE102017216536B4\_0061.png,DE102017216536B4\_0062.png"\; state="\{\{state\}\}">p</figure-callout>}}_y\cdot I_1+{\mathrm{<figure-callout\; id="2"\; label="p\; y"\; filenames="DE102017216536B4\_0058.png"\; state="\{\{state\}\}">p</figure-callout>}}_y\cdot I_2+p_{y3}\cdot I_3.\end{array}$$

und der x-Achse: $$\text{sin}\left({\alpha }_I\right)=\frac{\text{sin}\left(\alpha \right)+y_I}{\sqrt{{\left(\text{sin}\left(\alpha \right)+y_I\right)}^2+{\left(\text{cos}\left(\alpha \right)+x_I\right)}^2}}$$

$$\text{cos}\left({\alpha }_I\right)=\frac{\text{cos}\left(\alpha \right)+\Delta x_I}{\sqrt{{\left(\text{sin}\left(\alpha \right)+y_I\right)}^2+{\left(\text{cos}\left(\alpha \right)+x_I\right)}^2}}.$$

The angle a _I is an angle between the point $$P_I=\left(\begin{array}{c}\text{cos}\left(a\right)+x_I\\ \text{sin}\left(a\right)+y_I\end{array}\right)$$

and the x-axis: $$\text{sin}\left(a_I\right)=\frac{\text{sin}\left(a\right)+y_I}{\sqrt{{\left(\text{sin}\left(a\right)+y_I\right)}^2+{\left(\text{cos}\left(a\right)+x_I\right)}^2}}$$

$$\text{cos}\left(a_I\right)=\frac{\text{cos}\left(a\right)+\Delta x_I}{\sqrt{{\left(\text{sin}\left(a\right)+y_I\right)}^2+{\left(\text{cos}\left(a\right)+x_I\right)}^2}}.$$

und $$\begin{array}{l}\text{cos}\left(2\cdot {\alpha }_I\right)=\text{cos}\left({\alpha }_I+{\alpha }_I\right)\\ =\text{cos}\left({\alpha }_I\right)\cdot \text{cos}\left({\alpha }_I\right)-\text{sin}\left({\alpha }_I\right)\cdot \text{sin}\left({\alpha }_I\right)\\ =2\cdot \frac{{\left(\text{cos}\left(\alpha \right)+x_I\right)}^2\cdot {\left(\text{sin}\left(\alpha \right)+y_I\right)}^2}{{\left(\text{sin}\left(\alpha \right)+y_I\right)}^2\cdot {\left(\text{cos}\left(\alpha \right)+x_I\right)}^2}.\end{array}$$

A doubling of the angle a _I can be written as $$\begin{array}{l}\text{sin}\left(2\cdot a_I\right)=\text{sin}\left(a_I+a_I\right)\\ =\text{sin}\left(a_I\right)\cdot \text{cos}\left(a_I\right)+\text{cos}\left(a_I\right)\cdot \text{sin}\left(a_I\right)\\ =2\cdot \frac{\left(\text{sin}\left(a\right)+y_I\right)\cdot \left(\text{cos}\left(a\right)+x_I\right)}{{\left(\text{sin}\left(a\right)+y_I\right)}^2\cdot {\left(\text{cos}\left(a\right)+x_I\right)}^2}\end{array}$$

and $$\begin{array}{l}\text{cos}\left(2\cdot a_I\right)=\text{cos}\left(a_I+a_I\right)\\ =\text{cos}\left(a_I\right)\cdot \text{cos}\left(a_I\right)-\text{sin}\left(a_I\right)\cdot \text{sin}\left(a_I\right)\\ =2\cdot \frac{{\left(\text{cos}\left(a\right)+x_I\right)}^2\cdot {\left(\text{sin}\left(a\right)+y_I\right)}^2}{{\left(\text{sin}\left(a\right)+y_I\right)}^2\cdot {\left(\text{cos}\left(a\right)+x_I\right)}^2}.\end{array}$$

$$\Delta \text{ cos}\left(2{\alpha }_I\right)=\underset{t_{cx}}{\underbrace{\frac{d\text{ cos}\left(2{\alpha }_I\right)}{d{x}_I}}}\cdot \Delta x_I+\underset{t_{cy}}{\underbrace{\frac{d\text{ cos}\left(2{\alpha }_I\right)}{d{y}_I}}}\cdot \Delta y_I$$

$$\left(\begin{array}{c}\Delta \text{ sin}\left(2{\alpha }_I\right)\\ \Delta \text{ cos}\left(2{\alpha }_I\right)\end{array}\right)=\underset{T^T}{\underbrace{\left[\begin{array}{cc}{t}_{sx}& t_{sy}\\ t_{cx}& t_{cy}\end{array}\right]}}\cdot \left(\begin{array}{c}\Delta x_I\\ \Delta y_I\end{array}\right)$$

$$s:\text{sin}\left(\alpha \right)$$

$$c:=\text{cos}\left(\alpha \right)$$

$$t_{sx}=2\cdot \frac{\left(s+y_I\right)\cdot \left(s+y_I-c-x_I\right)\cdot \left(s+y_I+c+x_I\right)}{{\left({\left(s+y_I\right)}^2+{\left(c+x_I\right)}^2\right)}^2}$$

$$t_{sy}=2\cdot \frac{\left(s+x_I\right)\cdot \left(s+y_I-c-x_I\right)\cdot \left(s+y_I+c+x_I\right)}{{\left({\left(s+y_I\right)}^2+{\left(c+x_I\right)}^2\right)}^2}$$

$$t_{cx}=4\cdot \frac{\left(s+y_I\right)\cdot {\left(s+y_I\right)}^2}{{\left({\left(s+y_I\right)}^2+{\left(c+x_I\right)}^2\right)}^2}$$

$$t_{cy}=-4\cdot \frac{\left(s+{\text{x}}_I\right)\cdot \left(s+y_I\right)}{{\left({\left(s+y_I\right)}^2+{\left(c+x_I\right)}^2\right)}^2}.$$

T ist dabei eine Matrix, um den Einfluss der einzelnen Ströme auf die Sensorsignale zu gewichten.The influence of the errors x _I and x _I in relation to the measured sensor signals U _sin and U _cos can be related as follows: $$\Delta \text{sin}\left(2a_I\right)=\underset{t_{sx}}{\underbrace{\frac{\mathrm{i.e}\text{sin}\left(2a_I\right)}{\mathrm{i.e}{x}_I}}}\cdot \Delta x_I+\underset{t_{sy}}{\underbrace{\frac{\mathrm{i.e}\text{sin}\left(2a_I\right)}{\mathrm{i.e}{y}_I}}}\cdot \Delta y_I$$

$$\Delta \text{cos}\left(2a_I\right)=\underset{t_{cx}}{\underbrace{\frac{\mathrm{i.e}\text{cos}\left(2a_I\right)}{\mathrm{i.e}{x}_I}}}\cdot \Delta x_I+\underset{t_{cy}}{\underbrace{\frac{\mathrm{i.e}\text{cos}\left(2a_I\right)}{\mathrm{i.e}{y}_I}}}\cdot \Delta y_I$$

$$\left(\begin{array}{c}\Delta \text{sin}\left(2a_I\right)\\ \Delta \text{cos}\left(2a_I\right)\end{array}\right)=\underset{T^T}{\underbrace{\left[\begin{array}{cc}{t}_{sx}& t_{sy}\\ t_{cx}& t_{cy}\end{array}\right]}}\cdot \left(\begin{array}{c}\Delta x_I\\ \Delta y_I\end{array}\right)$$

$$s:\text{sin}\left(a\right)$$

$$c:=\text{cos}\left(a\right)$$

$$t_{sx}=2\cdot \frac{\left(s+y_I\right)\cdot \left(s+y_I-c-x_I\right)\cdot \left(s+y_I+c+x_I\right)}{{\left({\left(s+y_I\right)}^2+{\left(c+x_I\right)}^2\right)}^2}$$

$$t_{sy}=2\cdot \frac{\left(s+x_I\right)\cdot \left(s+y_I-c-x_I\right)\cdot \left(s+y_I+c+x_I\right)}{{\left({\left(s+y_I\right)}^2+{\left(c+x_I\right)}^2\right)}^2}$$

$$t_{cx}=4\cdot \frac{\left(s+y_I\right)\cdot {\left(s+y_I\right)}^2}{{\left({\left(s+y_I\right)}^2+{\left(c+x_I\right)}^2\right)}^2}$$

$$t_{cy}=-4\cdot \frac{\left(s+{\text{x}}_I\right)\cdot \left(s+y_I\right)}{{\left({\left(s+y_I\right)}^2+{\left(c+x_I\right)}^2\right)}^2}.$$

T is a matrix to weight the influence of the individual currents on the sensor signals.

$${\tilde{U}}_{cos}=A\cdot \text{cos}\left(2\alpha +\epsilon \right).$$

Estimation of an offset error can be estimated by the following assumptions, for example. An offset error is created by a misalignment between the permanent magnet and the sensor. The sensor model considering only one offset error ε can consequently be written as: $${\tilde{u}}_{sin}=A\cdot \text{sin}\left(2a+e\right)$$

$${\tilde{u}}_{cOs}=A\cdot \text{cos}\left(2a+e\right).$$

The offset error ε can also be modulated by: $$\begin{array}{l}e={\text{E}}_{\text{s1}}\text{sin}\left(a\right)+{\text{E}}_{\text{s}2}\text{sin}\left(2a\right)+{\text{E}}_{\text{s}4}\text{sin}\left(4a\right)+{\text{E}}_{\text{s}\mathrm{8th}}\text{sin}\left(\mathrm{8th}a\right)\\ +{\text{E}}_{\text{c1}}\text{cos}\left(a\right)+{\text{E}}_{\text{c2}}\text{cos}\left(2a\right)+{\text{E}}_{\text{c4}}\left(4a\right)+{\text{E}}_{\text{c8}}\text{cos}\left(\mathrm{8th}a\right).\end{array}$$

The actual rotor angle and/or the actual angular velocity and/or the actual angular acceleration is/are preferably derived from the estimated rotor angle error.

In particular, by considering an estimated rotor angle error, the rotor angle α, the angular velocity α′ and the angular acceleration α″ can be appropriately determined. In particular, the proposed method includes taking into account implicit functions, ie implicit dependencies, so that applicable values with regard to α, α′ and α″ can be determined.

The estimated rotor angle error from an earlier method step or is preferred from the same method step for calculating the rotor angle to be expected from the received angle signal, taking into account the estimated current error and/or the offset error.

The proposed method, which is carried out in particular during operation of the electrical machine, preferably comprises at least one loop in which an already estimated rotor angle is used as a value, in particular as a starting value, for the continuous estimation of the rotor angle to be expected. Particularly in the case of cyclic dependencies in the proposed method, which is preferably designed as an algorithm, the estimated rotor angle error from an earlier method step is used to calculate the rotor angle to be expected from the newly received angle signal. The use of estimated rotor angle errors from an earlier method step depends in particular on the implementation and leads to faster convergence of the proposed method.

According to the invention, an amplitude error and/or an offset error is/are also taken into account for calculating a rotor angle to be expected, in particular transferred to the calculation program.

$${\tilde{U}}_{cos}={\tilde{A}}_{cos}\cdot \text{cos}\left(2\alpha \right)+{\tilde{O}}_{cos}$$

wobei ein ideales Verhalten des Sensors angenommen wird, so dass $${\tilde{A}}_{sin}=A$$

$${\tilde{O}}_{sin}=0$$

$${\tilde{A}}_{cos}=A$$

$${\tilde{O}}_{cos}=\mathrm{0,}$$

The estimation of an amplitude error and an offset error can be estimated by the following assumption, for example. $${\tilde{u}}_{sin}={\tilde{A}}_{sin}\cdot \text{sin}\left(2a\right)+{\tilde{O}}_{sin}$$

$${\tilde{u}}_{cOs}={\tilde{A}}_{cOs}\cdot \text{cos}\left(2a\right)+{\tilde{O}}_{cOs}$$

where an ideal behavior of the sensor is assumed, such that $${\tilde{A}}_{sin}=A$$

$${\tilde{O}}_{sin}=0$$

$${\tilde{A}}_{cOs}=A$$

$${\tilde{O}}_{cOs}=\mathrm{0,}$$

$$\Delta U_{sin}=\underset{t_{sA}}{\underbrace{\frac{d{U}_{sin}}{d{A}_{sin}}}}\cdot \Delta A_{sin}+\underset{t_{sO}}{\underbrace{\frac{d{U}_{sin}}{d{A}_{sin}}}}\cdot \Delta O_{sin}$$

$$t_{sA}=\text{sin}\left(2\alpha \right)$$

$$t_{sO}=1$$

$$T_{sin}=\left[\begin{array}{c}sin\left(2\alpha \right)\\ 1\end{array}\right]$$

und $$\Delta U_{cos}=U_{cos}-{\tilde{U}}_{cos}$$

$$\Delta U_{cos}=\frac{d{U}_{cos}}{d{A}_{cos}}\cdot \Delta A_{cos}+\frac{d{U}_{cos}}{d{O}_{cos}}\cdot \Delta O_{cos}$$

$$t_{cA}=\text{cos}\left(2\alpha \right)$$

$$t_{cO}=1$$

$$T_{cos}=\left[\begin{array}{c}\text{cos}\left(2\alpha \right)\\ 1\end{array}\right].$$

A deviation of the parameters influences the difference between the estimated and the measured sensor signals in the following way: $$\Delta u_{sin}=u_{sin}-{\tilde{u}}_{sin}$$

$$\Delta u_{sin}=\underset{t_{sA}}{\underbrace{\frac{\mathrm{i.e}{u}_{sin}}{\mathrm{i.e}{A}_{sin}}}}\cdot \Delta A_{sin}+\underset{t_{sO}}{\underbrace{\frac{\mathrm{i.e}{u}_{sin}}{\mathrm{i.e}{A}_{sin}}}}\cdot \Delta O_{sin}$$

$$t_{sA}=\text{sin}\left(2a\right)$$

$$t_{sO}=1$$

$$T_{sin}=\left[\begin{array}{c}sin\left(2a\right)\\ 1\end{array}\right]$$

and $$\Delta u_{cOs}=u_{cOs}-{\tilde{u}}_{cOs}$$

$$\Delta u_{cOs}=\frac{\mathrm{i.e}{u}_{cOs}}{\mathrm{i.e}{A}_{cOs}}\cdot \Delta A_{cOs}+\frac{\mathrm{i.e}{u}_{cOs}}{\mathrm{i.e}{O}_{cOs}}\cdot \Delta O_{cOs}$$

$$t_{cA}=\text{cos}\left(2a\right)$$

$$t_{cO}=1$$

$$T_{cOs}=\left[\begin{array}{c}\text{cos}\left(2a\right)\\ 1\end{array}\right].$$

At least one observer is preferably used to determine the rotor angle error, which observer reconstructs non-measurable variables from known input variables, in particular disturbance variables, and output variables, in particular measured variables, of an observed system.

In control engineering, the term observer means a system that determines non-measurable variables from known input variables, such as manipulated variables or measurable disturbance variables, and output variables, such as measured variables such as the angle sensor signal, of an observed reference system. Here, the observer reproduces the observed reference system as a model and uses a controller to track the measurable, and therefore comparable with the reference system, measured variables. This can prevent a model from generating an error that grows over time, particularly in the case of reference systems with integrating behavior. An observer can be structured or set up in particular if the reference system can be observed using the available measured variables. Therefore, establishing observability using criteria is a necessary prerequisite for using an observer. Observers are used in measurement technology as a substitute for technically impossible measurements, such as in electrical machines.

A Luenberger observer or a Kalman filter is particularly preferably used as the observer.

The Kalman filter is a closely related alternative to the Luenberger observer. The Luenberger observer constantly amplifies the error, i.e. the difference between the estimated and measured sensor values. The theory behind a Luenberger observer is based on a parallel connection of the observer to a controlled system model. The difference between the sensor reading of the route, i.e. the received angle reading α, and a
"measured value" of the observer, i.e. an estimated value of the angle α̃, traced back to the model. This allows the observer to react to disturbances or their own inaccuracies. An advantage of the Luenberger observer is that the Luenberger observer can react to its own inaccuracies. In each step, the Kalman filter calculates a new gain based on assumptions about the noise of the sensors. This calculation also includes a matrix inversion. The Kalman filter thus requires more knowledge about the system, especially about the noise quantities, and tends to require more computing power for the matrix inversion. The advantage of using the Kalman filter can be that the estimated value converts to the true value more quickly.

An error from each of three current phases of the electrical machine is preferably taken into account for estimating the current error.

Furthermore, in one embodiment of the proposed method, it is assumed that the errors from the three current phases add up to an error. In this way, a current error can be passed on to the observer used.

To estimate the offset error, a difference is preferably calculated from the received angle signals with current errors.

Such a calculation can be performed using the equations already described herein.

According to a further aspect of the present invention, a microcontroller is proposed for calculating disturbances in a measured angle signal of a magnetic angle sensor of an electrical machine, the microcontroller being designed to receive the measured angle signal and to carry out a method as described herein.

According to a further aspect of the present invention, an electrical machine with a magnetic angle sensor and with a microcontroller described herein is proposed, the electrical machine having a rotor and a magnet arranged on the rotor.

Here, the magnet arranged on the rotor induces a measurable magnetic field in the angle sensor due to the movement of the rotor. The angle sensor preferably has a permalloy alloy or the like, in which a resistance can be measured as a function of a changing induced magnetic field. A permanent magnet is also arranged on the rotor, which induces the magnetic field in the angle sensor when the rotor moves.

According to a further aspect of the present invention, a computer program product is proposed which, when executed on a processor, causes the processor to execute a method described herein.

The invention described herein can be used above all in electrical machines of all types in which the position of the rotor is measured taking into account generated magnetic fields. As described here, the angle processing and the compensation of typical errors such as offset errors and/or amplitude errors of the sine and cosine signals as well as misalignments between the sensor and magnet and the compensation of the interference fields caused by the currents are performed by observers, in particular by a Luenberger observer or by a Kalman -Filter, realized.

A particular advantage of the embodiments of the method described herein is that the method or the various embodiments of the method do not require any outlay on parameterization. Furthermore, there is no need for magnetic shielding, which is usually expensive. The method described is therefore simpler and cheaper. In particular, the proposed method is self-learning, i.e. in the course of the application of the method, more and more accurate error values of the different types are estimated by the method.

Further advantages and features of the present invention result from the following description of preferred embodiments with reference to the attached figures. It goes without saying that individual embodiments shown in the respective figures can have features that can also be used in other embodiments, even if this is not explicitly mentioned and unless this was explicitly excluded due to technical circumstances. Show it:

• 1 eine Anordnung eines Sensors und eines Permanentmagneten an einem Rotor einer elektrischen Maschine,
• 2 eine schematische Darstellung des magnetoresistiven Effekts in Permalloy,
• 3 schematische Darstellung der Struktur eines Beobachters,
• 4 schematische Darstellung eines typischen Stromfehlers,
• 5 Schema eines Offset- und Amplituden-Beobachters,
• 6 Messergebnisse des Versatzfehlers unter Verwendung einer Ausführungsform des vorgeschlagenen Verfahrens,
• 7 Schema eines Versatzfehler-Beobachters,
• 8 Schema eines Stromfehler-Beobachters,
• 9 schematische Darstellung der Struktur des vorgeschlagenen Verfahrens gemäß einer bevorzugten Ausführungsform,
• 10 Messergebnisse, welche mit einem aus dem Stand der Technik bekannten Verfahren erhalten wurden, und
• 11 Messergebnisse, welche mit dem vorgeschlagenen Verfahren gemäß einer bevorzugten Ausführungsform erhalten wurden

The figures show:

• 1 an arrangement of a sensor and a permanent magnet on a rotor of an electrical machine,
• 2 a schematic representation of the magnetoresistive effect in permalloy,
• 3 schematic representation of the structure of an observer,
• 4 schematic representation of a typical current error,
• 5 Scheme of an offset and amplitude observer,
• 6 Measurement results of the offset error using an embodiment of the proposed method,
• 7 scheme of an offset error observer,
• 8th Schematic of a current error observer,
• 9 schematic representation of the structure of the proposed method according to a preferred embodiment,
• 10 Measurement results obtained using a method known from the prior art, and
• 11 Measurement results obtained with the proposed method according to a preferred embodiment

Further advantages of popular embodiments are summarized below 1 until 11 explained, especially in comparison to the methods already known and used from the prior art.

1 shows an arrangement of a magnetic angle sensor 12 and a permanent magnet 14, which is arranged on a rotor 16 of an electrical machine 10. As in 1 the sensor 12 is shown opposite the permanent magnet 14 on a substrate 17 . The substrate 17 is preferably designed as a PCB board or as a ceramic board. The sensor 12 is arranged opposite the permanent magnet 14 along an axial axis 18 which runs along an extension of the rotor 16 . A position of the rotor 16 on which the permanent magnet 14 is arranged can be measured using the magnetic angle sensor 12 . The magnetic angle sensor 12 can comprise a permalloy alloy, for example, in which the resistance changes with the direction of the magnetic field. A movement 20 of the rotor 16 is in 1 indicated for example by the arrow 20. The movement 20 of the rotor 16 can consequently be detected by the angle sensor 12 . The angle sensor 12 detects, for example, the rotor position angle α, the rotor angular speed α′ and/or the rotor angular acceleration α″.

The magnetic angle sensor 12 preferably has an alloy in which the measurable resistance changes in the form of a magnetization M as a function of a direction of a magnetic field H. This effect is known as the magnetoresistive effect. Such an alloy can be a permalloy alloy, for example. A permalloy alloy preferably has up to 80% by weight nickel and up to 20% by weight iron. 2 shows a schematic representation of the magnetoresistive effect. In this case, I indicates a current which can be measured in the alloy 20 and which can be measured between a voltage difference 22 , 24 . It is also conceivable that an alloy other than permalloy is used as long as the alloy used exhibits the magnetoresistive effect.

3 shows a schematic representation of the structure of an observer. The sensor 12 records the values (sin(θ), cos(θ)) and transmits them to a calculation model. A is the system matrix and C the output matrix of the system (A,C). The quantities A and C can be used to determine whether the system is observable. The system is observable if, for example, the Kalman criterion or another criterion is met. If this is the case, a state E, of the system can be determined, with the estimation error ϑ - ϑ̃ being amplified with the matrix L in order to correct the estimated value for the state E, of the system. If the matrix (A-LC) has only eigenvalues with negative real part, the observer is stable. The system of equations is then solved by integration, whereby the system state E, is determined.

4 shows a schematic representation of a typical current error X in vectorial form. A current I measured by the angle sensor 12 includes the current error components X=(x _I , y _I ), which lead to the measured angle α _I . The measured phases of the current themselves are usually not faulty. However, the magnetic field caused by the phase currents causes an error in the sensor. The measured angle α _I therefore does not correspond to the actual angle α, so that the measured angle α _I has to be corrected accordingly by subtraction.

5 shows a schematic of an offset and/or amplitude observer. The variable ΔU is the difference between the measured sensor value (sin(θ), cos(θ)) and the estimate. The variable L is a freely selectable error amplification. T is the weighting of the error, which is given by a partial derivative of the sensor signal according to the error sources amplitude and offset. The estimates for amplitude and offset are adapted by integrating the sensor error - weighted with L and T.

The influence of the offset error is usually the same, while the influence of the amplitude error changes with the angle α. In this way, a distinction can be made between the error, which includes the offset, and the error, which includes the amplitude error.

6 shows measurement results of offset error versus rotor angular position using an embodiment of the proposed method. 6a 12 shows measured sensor signals measured by sensor 12 and ideal cosine and sine curves without an offset error model for comparison. The 6b shows a resulting angular error AgErr resulting from the offset error superimposed with a modeled angular error AgErrMdl. 6c shows the difference between the angle error AgErr minus the modeled angle error AgErrMdl. In particular the 6c a constant value can be taken from it, which shows a fluctuation of less than 1 degree in absolute terms.

7 shows a scheme of an offset observer, where the offset error is estimated as a difference between the measured angle and the estimated angle. Otherwise, the calculation runs parallel to the calculation described 5 , which is hereby incorporated by reference.

In contrast to the 5 and 7 shows 8th a scheme of a current error observer. This differs further in that an error is associated with each phase I ₁ , I ₂ , I ₃ of current I, so that the current error components x _I and y _I can be written as a sum, as described herein. The parameters of the matrix T are freely selectable and include amplification factors $$L=\left[\begin{array}{c}{I}_1\\ I_2\\ I_3\end{array}\right].$$

With the detected current of each phase, the parameters p _xn ,p _yn, n= 1,2,3 can be adapted accordingly, as is shown schematically in 8th is shown.

9 shows a schematic representation of the structure of the proposed method according to a preferred embodiment. In this procedure the 9 Among other things, the compensation of interference in a measured angle signal of a magnetic angle sensor of an electrical machine takes place. The method has the following steps: receiving a measured angle signal, estimating a current error and/or an offset error of the measured angle signal, and calculating a rotor angle to be expected from the measured angle signal, taking into account the estimated current error and/or the estimated offset error, in particular during operation the electric machine.

The 9 shows an overview of the signal flow of the proposed method. The idea of the method or the algorithm is to determine non-measurable variables by comparing expected sensor signals (sine, cosine signals) with actually measured sensor signals. This happens in step 30 'Rotor Angle Error Calculation'. Before that, however, sensor signals measured by the angle sensor 12 in step 29 are transferred to the algorithm described. The estimate for the rotor state (angle, angular velocity, angular acceleration) is improved by means of the angular error. This happens in step 32 'Rotor Shaft State Observer'. The expected sensor values can be calculated from the estimated rotor angle with ideal values for the amplitude and the offset of the sensor. This happens in step 34 'Sensor Model'. If the sensor behavior deviates from the ideal, this affects the estimation of the rotor state. In a step 36, a loop 36, the calculated sensor values can be transferred again to the calculation method for compensating for interference fields until no or only a small deviation from the ideal can be determined. At the same time, in a step 31, the detected sensor signals are transferred directly to the sensor model 34, so that the sensor model can determine the parameters for the sensor model from the measured and the estimated values.

10 shows model results obtained with a method known from the prior art. The estimation according to the method known from the prior art leads to strongly scattering values between 14800 and 15100 rpm, ie to values which scatter over a range of 300 rpm.

11 on the other hand, model results are shown which were obtained with the proposed method according to a preferred embodiment. With the proposed method, values are achieved which scatter constantly in a small range of up to 20 rpm.

The curves 40 show in the 10 and 11 respectively measured curves or results and the curves 42 respectively calculated curves.

Reference List

10

electric machine

12

magnetic angle sensor

14

permanent magnet

16

rotor

17

substrate

18

rotor axis

20

movement of the rotor

21

alloy

22, 24

potential difference

I

Electricity

H

magnetic field

M

magnetization

a

rotor angle

α'

angular velocity

α''

angular acceleration

29

Step

30

Step rotor angle error calculation

32

Step rotor condition observer

34

sensor model

36

Ribbon

40

measured curve

42

calculated curve

Claims (12)

Method for compensating for disturbances in a measured angle signal of a magnetic angle sensor (12) of an electrical machine (10), the method having: - receiving a measured angle signal, - estimating a current error and/or an offset error of the measured angle signal, wherein for estimating the current error and/or the offset error, the measured angle signal, which comprises at least two components, is transferred to a calculation program stored in a microcontroller, - calculating a rotor angle to be expected from the measured angle signal, taking into account the estimated current error and/or the estimated offset error, during operation of the electrical machine (10), characterized in that an amplitude error and/or an offset error is also taken into account to calculate a rotor angle to be expected, in particular is transferred to the calculation program. procedure after claim 1 , wherein the calculation program comprises a model with which a rotor state detected by means of the received angle signal comprising an angle and an angular velocity and an angular acceleration, taking into account various errors in the angle signal comprising an amplitude error and/or an offset error and the offset error and/or the current error on one current actual rotor state is mapped. Procedure according to one of Claims 1 - 2 , wherein the estimated current error and/or the estimated offset error is used to estimate a rotor angle error by determining a difference of a measured rotor angle minus an estimated rotor angle. procedure after claim 3 , wherein the actual rotor angle and/or the actual angular velocity and/or the actual angular acceleration is/are derived from the estimated rotor angle error. Procedure according to one of claims 3 or 4 , wherein the estimated rotor angle error from an earlier method step or from the same method step is used to calculate the rotor angle to be expected from the received angle signal, taking into account the estimated current error and/or the offset error. Procedure according to one of Claims 1 - 5 At least one observer is used to determine the rotor angle error, which observer reconstructs non-measurable variables from known input variables, in particular disturbance variables, and output variables, in particular measured variables, of an observed system. procedure after claim 6 , using a Luenberger observer or a Kalman filter. Procedure according to one of Claims 1 - 7 , wherein an error from each of three current phases of the electrical machine (10) is taken into account for estimating the current error, and it is assumed in the method that the errors from the three current phases add up to form an error. Procedure according to one of Claims 1 - 8th , wherein a difference is calculated from the received angle signals with current errors to estimate the offset error. Microcontroller for calculating disturbances in a measured angle signal of a magnetic angle sensor (12) of an electrical machine (10), the microcontroller being designed to receive the measured angle signal and a method according to one of Claims 1 - 9 to execute. Electrical machine with a magnetic angle sensor (12) and a microcontroller claim 10 , wherein the electrical machine (10) has a rotor (16) and a permanent magnet (14) arranged on the rotor (16), the permanent magnet (14) generating a measurable magnetic field in the rotor (16) by a movement (20) of the rotor Angle sensor (12) induced. A computer program product which, when executed on a processor, causes the processor to perform a method according to any one of Claims 1 - 9 to execute.

Images